The reemployment bonus is a cash payment made to an Unemployment Insurance (UI) recipient who finds a job within a relatively short period after filing for UI benefits. Two randomized trials of the reemployment bonus have already been completed, one in Illinois (Woodbury and Spiegelman 1987), the other in New Jersey (Mathematica Policy Research 1989), and both have suggested that a reemployment bonus program could substantially reduce the duration of insured unemployment without adverse consequences for workers covered by the bonus program. Additional, more refined, experiments with the reemployment bonus are currently in progress in Washington State and Pennsylvania.

Nevertheless, an important concern surrounding the reemployment bonus is that it could reduce the number of steady-state jobs held by workers who are not covered by the program. To take the extreme case, if each new job obtained by a covered worker were at the expense of an uncovered worker, then the program's effect would simply be distributional and there would be no real effect on total unemployment. This "displacement effect," if large, would greatly decrease the attractiveness of the reemployment bonus.

The purpose of this paper is to develop a simple partial equilibrium matching model of the labor market that allows us to investigate the nature and size of the displacement effect. We accomplish this by assuming that it takes time and effort for unemployed workers and firms with vacancies to find each other. Workers can increase the probability of finding employment by increasing their search effort, but increased search effort is costly. In equilibrium, workers choose a level of search activity that equates the expected gain from additional search with marginal cost. The bonus program increases the expected payoff to search for covered workers and therefore results in an increase in their search effort. This increase in search activity has three effects. First, if we hold search effort by uncovered workers constant, then there will be an immediate increase in the number of jobs held by covered workers. Some of these new jobs may come at the expense of uncovered workers but others will simply be due to the fact that greater search activity allows the economy to make better use of the existing search technology and therefore operate closer to full employment. The new jobs created will eventually benefit uncovered workers as well since when the worker needs to be replaced, both covered and uncovered workers will be free to compete for the job opening. We refer to the increase in overall employment due to the change in the covered workers' search effort as the "gross job creation effect" of the bonus program. The fact that some of the increase in covered employment comes at the expense of uncovered workers is referred to as the "direct substitution effect."

The third effect of the bonus program is generated by the change in the search behavior of uncovered workers. Since the increased search effort of covered workers will make it more difficult for uncovered workers to find employment, the bonus program will trigger an increase in search effort of uncovered workers as well. This "rivalry effect" will increase the number of jobs held by uncovered workers and will tend to neutralize the direct substitution effect. The result is increased employment for both covered and uncovered workers. Moreover, even in cases in which the direct substitution effect dominates, we find that the displacement effect is relatively small.

The paper divides into three additional sections. In the first section we introduce a simple version of a partial equilibrium matching model that is patterned after the work of Diamond (1982), Mortensen (1982) and Pissarides (1984). This model incorporates the elements of the Illinois reemployment bonus experiment, which offered a $500 cash bonus to insured workers who were reemployed within 11 weeks of filing for UI. In this simple version of the model, wages are assumed to be exogenous and unaffected by the program (this is consistent with empirical evidence gathered during the Illinois experiment). Data gathered in the course of evaluating the experiment are then used to infer values for the unobservable parameters of the model. Finally, estimates of the displacement effect are obtained by solving the model assuming that the program is in effect. In the second section, the model is extended to allow wages to be determined endogenously. We demonstrate that although this complicates the analysis considerably, it does not affect that qualitative nature of our results. That is, we show that the result that the displacement effect is small (or non-existent) is robust. In the final section, we begin by summarizing our results. We then go on to argue that while our model is rich enough to capture many of the essential features of frictional unemployment, it is also simple enough to be used to study the displacement effects of a variety of government programs. The analysis provided in this paper serves as an example of its potential value.
NOTE: A revised version of this paper appears in Journal of Labor Economics, Vol. 11, 575-605, October 1993.